Interwell Connectivity and Diagnosis Using Correlation of Production and Injection Rate Data in Hydrocarbon Production

摘要

This report details progress and results on inferring interwell communication from well rate fluctuations. Starting with the procedure of Albertoni and Lake (2003) as a foundation, the goal of the project was to develop further procedures to infer reservoir properties through weights derived from correlations between injection and production rates. A modified method, described in Yousef and others (2006a,b), and herein referred to as the 'capacitance model', is the primary product of this research project. The capacitance model (CM) produces two quantities, {lambda} and {tau}, for each injector-producer well pair. For the CM, we have focused on the following items: (1) Methods to estimate {lambda} and {tau} from simulated and field well rates. The original method uses both non-linear and linear regression and lacks the ability to include constraints on {lambda} and {tau}. The revised method uses only non-linear regression, permitting constraints to be included as well as accelerating the solution so that problems with large numbers of wells are more tractable. (2) Approaches to integrate {lambda} and {tau} to improve connectivity evaluations. Interpretations have been developed using Lorenz-style and log-log plots to assess heterogeneity. Testing shows the interpretations can identify whether interwell connectivity is controlled by flow through fractures, high-permeability layers, or due to partial completion of wells. Applications to the South Wasson and North Buck Draw Fields show promising results. (3) Optimization of waterflood injection rates using the CM and a power law relationship for watercut to maximize economic return. Tests using simulated data and a range of oil prices show the approach is working. (4) Investigation of methods to increase the robustness of {lambda} and {tau} estimates. Human interventions, such as workovers, also cause rate fluctuations and can be misinterpreted by the model if bottom hole pressure data are not available. A revised method, called the 'segmented capacitance model', identifies times when production changes might not be caused strictly by water injection changes. Application to data from Monument Butte Field shows encouraging results. Our results show the CM and its modified forms can be an important tool for waterflood management. We have moved beyond the proof of principle stage to show it can actually be applied to assess connectivity in field situations. Several shortcomings, however, remain to be addressed before the CM can be routinely applied by field operators. The CM and its modifications analyze well rates in the time domain. We also explored the assessment of interwell connectivity in the spectral domain. We applied conventional methods, based on analyzing passive linear electrical networks, to the analysis of injection and production data. In particular, we assessed the effects of near-wellbore gas on the apparent connectivity. With only oil and water in the system, the results were as expected, giving good connectivity estimates. In the presence of gas, however, the methods could not produce useful estimates of connectivity.